Object Detection
A computer vision task that identifies and localizes multiple objects within an image by predicting bounding boxes and class labels simultaneously.
Object detection identifies multiple objects in an image by predicting both their spatial locations (bounding boxes) and category labels. While classification answers "what is in this image," detection answers "where is each object and what is it."
The field spans autonomous driving, surveillance, robotics, and medical imaging. Accuracy and speed trade off against each other.
- Two-stage detectors: Exemplified by Faster R-CNN, these first generate region proposals then classify and refine each one. They achieve high accuracy but at greater computational cost
- One-stage detectors: YOLO and SSD predict boxes and categories directly with a single network, without a proposal stage. They enable real-time inference
- Transformer-based detectors: DETR reformulates detection as a set prediction problem using attention, eliminating Non-Maximum Suppression. RT-DETR is a DETR-family model aimed at real-time use
Evaluation uses mAP (mean Average Precision) computed across multiple IoU thresholds. Benchmark datasets include COCO.